Top Ten Best Practices for Data Integration
Use these guidelines to help you achieve more modern, high-value and diverse uses of DI tools and techniques.
It'd be nice to think that, as enterprises move to the cloud, they're learning from their past mistakes, particularly when it comes to integration.
Yes, it'd be nice, but, alas, that doesn't seem to be what's happening.
Ross Mason, CTO and founder of MuleSoft, writes that many enterprises are still using point-to-point integration, often built in-house, to connect with the cloud. You might recognize the criticisms since point-to-point is the same approach that's led to brittle, breaking connections within the enterprise.
And just like Rocky, Mason reminds us "That trick never works!" Granted, he may not have the gravitas of a flying squirrel, but he's still pretty convincing in his thesis against point-to-point integration with the cloud. Writing as a guest columnist for Dana Gardner of Interarbor Solutions, Mason lists 10 pitfalls you'll encounter by using point-to-point integration in the cloud.
And just like Bullwinkle reaching into that hat for a rabbit and pulling up an angry lion instead, these point-to-point integrations seem like no big deal. But before too long, you've got trouble with a capital "T."
For instance, let's say you fall for pitfall number two: a "quickfire" integration. You need two systems to connect, so a developer codes a connection in a matter of days. It goes well, so you decide to integrate a few more systems thinking it'll only take a few more days. Wrong. Point-to-point doesn't scale, so you've got a more complex project on your hands.
Other problems you may inadvertently create, according to Mason, are:
Of course, Mason has a vested reason to discourage home-style point-to-point integration. Mason's company, Mulesoft, offers an iPaaS solution called Mule iON. It's an integration platform as a service, which means it supports not just designing, but maintaining and supporting your integration. Basically, its "engine" for handling the integration work is Mule's ESB.
But just because Mulesoft offers an alternative solution doesn't mean Mason is wrong. You could argue it's why Mason's right. There's a reason ESBs have become so popular for integration in the enterprise, and it's because it scales well and is less brittle than hand-coded, point-point integration.
In fact, I think most integration experts would agree with all 10 of his pitfalls, since, by and large, data experts frown on point-to-point integration. John Schmidt, Informatica's vice president of Global Integration Services, even goes so far as to list point-to-point integration as number one among the Seven Deadly Sins of Integration.
Data virtualization, aka data services, also offers an alternative to point-to-point integration. Robert Eve, the EVP of marketing for Composite Software, recently named cloud integration as one of the emerging popular use patterns for data virtualization. Of course, Composite Software offers a data virtualization solution, and, true to his title, Eve sees data virtualization as "ideally architected to solve."